A human mesenchymal spheroid prototype to replace moderate severity animal procedures in leukaemia drug testing

Materials

Methods

Ethical approval

Patient leukaemia cells were obtained from the Newcastle Biobank (REC reference number 07/H0906/109+5). All samples were acquired following written informed consent. Animal data shown here reflect unpublished data that were existing from past experiments. These experiments were carried out as per previously published methods.^12,17,25 PDX-ALL samples used in this project are from our cryopreserved PDX bank where the samples were generated in previous/earlier projects. All animal studies were performed in accordance with UK Animals (Scientific Procedures) Act, 1986 under project licence P74687DB5 following approval of Newcastle University animal ethical review body (AWERB).

2D BM-MSC-ALL co-culture

Patient derived PDX-ALL samples were cultured on 2D monolayers of MSC cells using a protocol published in an earlier study.¹⁷ MSC used in this study are between passages ranging from p2 – p5. MSC were seeded on Matrigel coated 48 well plates at 1×10⁴ cells/0.5 mL/cm² in their respective media (MSC media, Table 1). After 24 hours, the media was aspirated, and the cells rinsed with 1 mL SFEM. Leukaemia cells were then plated onto the MSC layers, at a density of 0.25×10⁵ cells/1 mL per well, suspended in SFEM media.

3D culture of BM-MSC

MSC were transferred to a 6-well low adhesion plate and cultured in 2 mL MSC media (Table 1) for 48 hours, after which, MSC spheroids were seen to form. Spheroids obtained using this method were irregular in shape and therefore experiments were repeated and optimised using Aggrewell400 plates. Aggrewell plates were prepared by rinsing each well with 0.5 mL of Aggrewell rinsing solution, spinning at 1500 g for 5 minutes and then aspirating excess solution. MSC were seeded at varying concentrations to track effect of seeding density on spheroid size. Cells were trypsinised and seeded in single cell format on Aggrewell plates at varying seeding densities in MSC media. Aggrewell plates were spun at 100g for 3 minutes to ensure all cells reached the bottom of the microwells. After 48 hours, MSC spheroids were ready for harvest and manually transferred onto low adhesion plates. Using a stereomicroscope fitted within a Class II biological safety cabinet, the spheroids were transferred onto respective culture-ware using a Gilson p1000 pipette with the tip-end cut off using sterile scissors.

3D BM-MSC-ALL co-culture

Leukaemia cells from PDX-ALL samples were co-cultured with the MSC spheres in SFEM media, at 2×10⁵ cells/mL, in 5 ml final volume, using low adhesion 48 well plates over a 5-day period. On day of harvest, the leukaemia cells and MSC spheroids were transferred into a 30 ml universal tube. This was centrifuged at 1500 RPM for 5 minutes. After aspirating the supernatant, 0.5 ml of X1 Trypsin EDTA was added and cultures incubated at 37C, 5% CO₂ for 2 minutes. Cells were mechanically dissociated into single cells by resuspending with a pipette. 4 ml of FBS was added and this was centrifuged at 1500 RPM for 5 minutes. Supernatant was aspirated and cells were resuspended in 1ml of fresh SFEM media. During cell count leukaemia cells were distinguished from MSC based on their morphology and size difference, the MSC are 3-4 times the size of leukaemia cells.

Dexamethasone 3D drug assay

10 million leukaemia cells were spun at 500g X 5 minutes and plated onto 3D MSC spheres in 48 well plates as described above. 5 nM dexamethasone (stock dissolved in methanol to generate 10 mM dexamethasone concentration, working concentration of dexamethasone was made by performing serial dilutions using SFEM media) was added to desired wells and leukaemia cells were cultured for 5 days. Experiments were repeated to include a total of 3 biological replicates with each replicate including 3 independent experiments.

3D Spheroid Calcein AM/Hoechst-33342 staining

MSC were seeded into Aggrewell plates to obtain a final spheroid size of 200 cells per microwell. Cells were cultured in the Aggrewell format for 3 and 7 days following which resultant spheroids were collected for the staining procedure. The calcein stain was prepared by reconstituting the lyophilized reagent to a 1mM stock in DMSO. From this, 1 μM working solution was prepared by diluting into sterile calcium free PBS. The MSC spheres were resuspended in PBS at a density of approximately 1×10⁷cells/ml, to which 10 μl of 1 μM calcein stain was added to final concentration of 0.01 μM. Cells were incubated at 37°C for 20 minutes in the dark. Cells were pelleted and resuspended in pre warmed media and left for 10 minutes to ensure optimal calcein retention before imaging. Samples that were double stained with Hoechst-33342 were stained for calcein as above and following the 10 minute incubation, were stained with Hoechst-33342 at final working concentration of 10 μg/ml in pre warmed media for 20 minutes. Digital Images were captured using Leica Application Suite X (LASX) version 3.5.7.23225.

MSC Marker Antibody staining

MSC Spheroids were collected from Aggrewell plates, disrupted to single cells using a trypsin EDTA wash followed by gentle mixing. Remaining cell clumps were removed by passing sample through a 40 μm PluriStrainer. Cells were counted using a trypan blue exclusion assay. 1 million cells were resuspended in PBS to wash before blocking in 10%FBS/PBS. Cells were again spun down and resuspended in 200 μl PBS. 5 μl antibody was added and samples were incubated at room temperature for 1 hour in the dark. Following incubation, 800 μl 10% FBS/PBS was added and cells were spun down again before being resuspended in PBS for running on flow cytometer.

Extraction of mRNA

RNA was purified from cell pellets (viable cells centrifuged at 5000 g X 5 minutes followed by aspiration of supernatant media) using the RNeasy mini kit, and as per manufacturer protocols. The RNA yield and quality was checked using the nanodrop ND-1000 spectrophotometer.

cDNA synthesis

RevertAidTMH Minus First Strand cDNA Synthesis Kit was used to synthesise cDNA from the RNA isolated. 500 ng RNA was collected and added to RNase/DEPC free water to a final volume of 11 μl. 1 μl (dN)6 (200 mg/l) random hexamers was added, mixed gently by inverting the vial and briefly centrifuged. Using a GeneAmp PCR system 2700 the sample was incubated at 65°C for 5 minutes, after which the sample was immediately placed on ice, 8 μl of the master mix (Table 2) was added, the samples were vortexed and briefly centrifuged. The samples were placed back in the PCR machine to incubate at 25°C for 10 minutes, 42°C for 60 minutes and 75°C for 10 minutes to terminate the reaction. The product was either used immediately for RT-PCR or stored at -20°C for short term storage.

qRT-PCR

Upon receipt, primers were reconstituted in RNase/DNase free water to a final stock concentration of 100 μM. PCR Mix (Table 3) was loaded onto PCR plate in triplicates. The plate was sealed and centrifuged for 1 minute at 1000 RPM and placed in an applied Biosystems 7900HT Sequence Detection System. ViiA7TM System was used to run the qRT-PCR plate and cycle threshold (Ct) and melting curves were obtained for analysis.

Costings analysis of in vivo versus in vitro models

Costings of in vitro models were estimated using list prices of media, reagents and other materials associated with developing and using the models for a 2 drug combination dose finding assay. This included 5 biological replicates, 3 experimental repeats with each having 3 technical repeats with a total number of experiments of 45. These costs were divided into the following categories: Production of clinical/patient-derived leukaemia samples; in vitro CRISPR/RNAi organoid generation; Dose/IC50 finding for 2 drug combination; and evaluation of drug combination.

In vivo costings were calculated using figures obtained from the Comparative Biology Centre, Newcastle University. This included fixed costs including mouse purchase from an NSG in-house colony, housing and IVIS use. Comparative in vivo equivalents to the in vitro design accounted for: production of PDX engrafted tissue (5 PDX, 6 NSG mice/PDX); in vivo CRISPR/RNAi (control and treatment groups, 5 NSG mice each); Dose finding for 2 drug combination (3 mice per sex); and evaluation of the combination (4 arms, 6 mice per arm). All total costings were calculated using Microsoft Excel and figures were created using Microsoft Excel.

Data analysis

Microscopy images were analysed using the Windows download bundled with Java 8. Version 1.4.3. Real time qPCR data were captured and analysed using the StepOne Plus Real-Time PCR System (ThermoFisher Catalog No: 4376600) and Software package StepOne Software v23.

Development and validation of BM-MSC spheres

We find that 2D ex vivo human cell-based co-culture platforms potentially enable 73% (metrics obtained locally) replacement of animals in drug testing experiments. Drug testing experiments are conventionally carried out using PDX-mouse-models, currently regarded as gold standard preclinical leukaemia models (Figure 1). Here we develop 3D BM-MSC-spheroids, to improve the spatial biomimicry of functional cell-cell contacts, by subculturing MSC onto low adhesion non-tissue culture treated plates (Figure 2a-b). We show that 3D MSC retain expression of key genes such as THY1 (CD90), CDH2 and NES (Figure 2c).

Figure 1. Comparison of existing in vivo preclinical leukaemia models with newly developed ex vivo animal replacement organoid models.

Key benefits of the animal replacement organoid models include species specificity and consequently higher biomimicry.

Figure 2. Aggrewell plates generate uniform 3D spheroids from single cell suspension of the desired cell-type. SD from three independent experiments (using MSC from same patient) have been plotted as error bars. Phase contrast photographs showing (a) 2D MSC, scale bar = 100 μM and (b) 3D MSC, scale bar = 200 μM after 96 hours of growth. (c) Mesenchymal marker levels quantified through qPCR and normalised to GAPDH. (d,e) Phase contrast photograph of MSC spheroids generated through culture using Aggrewell plates. Scale bar = 1000 μM (f) Correlation between cell seeding densities and size of resultant spheroids. Data captured following 48 hours of seeding. (g) CD105, CD90 and CD73 expression in MSC. Black = unstained. Blue = 2D MSC. Red = 3D MSC. (h) Hoechst-33342-Calcein staining in 100 μM spheroids, taken at day five of 3D culture. Calcein (green) stains live cells, Hoechst-33342 (blue) stains live and dead cells.

Aggrewell400 plates are designed for higher throughput spheroid engineering. These plates further aid the production of MSC spheres that show greater consistency in shape and size. Each well on these plates house 400 conical microwells, funnelling single cells and encouraging them to conglomerate into spheroids (Figure 2d). Here we use these microwells to create uniform BM-MSC spheres (Figure 2e), consequently allowing standardisation of the 3D technique. We find a distinct correlation between seeding cell density and size of spheroids, with higher seeding densities resulting in generation of more uniformly sized spheroids (Figure 2f). For subsequent experiments, we use MSC spheroids of diameter 100 μM size, generated with 200 MSC seeded per microwell. In addition, we show that MSC harvested from these 3D spheroids retain the expression of MSC markers CD105, CD90 and CD73 at protein level (Figure 2g). We furthermore confirm the retention of cell viability within these spheres via Hoechst-33342 dye, which stains live and dead cells and calcein dye, which stains live cells (Figure 2h).

Patient-derived ALL cells co-cultured with BM-MSC show superior proliferation and reduced dexamethasone sensitivity

We co-culture PDX-ALL cells with 2D MSC or with 3D MSC-spheroids, and monitor ALL cell counts over a 5-day period. We find that when co-cultured with MSC-spheroids, the leukaemia cells show 2-fold higher cell proliferation compared to 2D MSC-co-cultures (Figure 3a-c). Next, we repeat the co-cultures under low dose dexamethasone (5nM) treatment pressure. Dexamethasone is a glucocorticoid routinely used in the clinics to treat ALL. We find that ALL cells show greater reduction in sensitivity to dexamethasone when co-cultured with 3D BM-MSC (Figure 3d-f). This could potentially be due to improved cell-cell contacts and physiologically relevant cell polarity being achieved in a 3D format. This might have caused increased 3D versus 2D MSC-conferred ALL proliferation and/or improved ALL survival. Reduced sensitivity could also be due to altered drug availability within 3D spheroids. These data highlight the advantage of 3D organoid-based co-culture systems over routine 2D cultures in two respects: 1. Achieving higher ALL cell cycling which is essential for the action or testing of majority of anti-cancer treatments 2. Reduced drug sensitivity, possibly owing to improved PDX-ALL survival via MSC-conferred treatment protection. We further investigate the ability of MSC-spheroids to reduce ALL cell death and find that when treated with a higher dose of dexamethasone, that is, 10nM, then both 2D and 3D MSC-spheroids support survival of leukaemia cells (Figure 3g). We now progress to reveal proof-of-confidence-in-application data, justifying the suitability of our platform in conducting combinatorial drug testing.

Figure 3. 3D MSC-spheroids confer higher proliferation and improved survival onto patient-derived ALL. Growth kinetics, measured via viable cell counts of ALL samples (a,d) L49120, (b,e) MS40 and (c,f) L707D across a 5 day period, with and without low dose dexamethasone (10nM) pressure, as co-cultures with 3D MSC spheres versus 2D MSC-co-culture. L49120, MS40 and L707D refer to three different patient samples, i.e., 3 biological repeats. (d,e,f) Dexamethasone treatment, data shown for 5nM treatment, is less effective in a 3D microenvironment. N= 3. (g) Flow cytometry data, representative plots, showing propidium iodide, PI, positive ALL cells (sample L707) following five days of 10nM dexamethasone treatment. The column graph shows % ALL cells stained positive for PI. N = 2.

Co-cultures of 3D BM-MSC with PDX-ALL cells how high in vivo predictivity

Improved in vivo predictivity is a key requisite for ex vivo animal replacement models, if these platforms are to effectively replace mouse experiments. We confirm that BM-MSC/leukaemia 2D co-culture platform generate drug dose response data that was largely comparable with drug response observed in vivo in PDX-mouse models (Figure 4a). Furthermore, we reveal that when PDX-ALL cells are co-cultured with 3D BM-MSC-spheroids they show greater reduction in sensitivity, compared to the 2D co-culture arm, against dexamethasone, ABT-199 (a molecularly targeted therapy against BCL-2 positive blood cancers), and dexamethasone-ABT-199 combination in a sample subgroup that exhibited reduced in vivo efficacy (obtained from previously unpublished in vivo data existing in the lab from earlier studies). This shows the ability of our 3D co-culture model to screen off false-positive hits derived from 2D models, consequently minimising in vitro to in vivo attrition rates. We further confirm that when co-cultured with 3D MSC spheres, ALL cells expectedly proliferate in 3D-MSC spheroid-co-cultures, however, fail to increase in cell counts on day five, in the dexamethasone treatment arm (Figure 4b). Moreover, we show that neither dexamethasone, nor ABT-199, have any effect on survival of MSC, when treated for five days within a 3D sphere format (Figure 4c). Finally, we perform a costs analysis and show improved financial sustainability, namely a 30% costs benefit, in our ex vivo organoid platform compared to animal models (Figure 5).

Figure 4. In vitro co-culture platform delivers drug response data with high in vivo predictivity to minimise in vitro to in vivo drug attrition rates. Leukaemia cells from leukaemia cytogenetic subgroup A is a responder to ABT199 (targets BCL2 in BCL2 positive ALL) and dexamethasone drug combination in vitro and in vivo. On the other hand, leukaemia cells from subgroup B show low sensitivity to this drug combination in vitro in both 2D and 3D formats, and subsequently poor response is also noted in vivo using PDX mice models. In vitro data readout includes viable cell counts; in vivo data (right hand graphs) is shown as bioluminescence imaging from luciferase expressing PDX cells in total flux(p/s). Total flux is a surrogate of blast number in vivo. Error bars shown refer to standard error (SE) using 4 mice per treatment group. Dexamethasone and ABT-199 are used in the clinics to treat ALL. b. ALL cell counts on day 0 versus day 5 of culture, with and without treatment with 25 nM dexamethasone. N = 3. Error bars = SD. c. Cell counts following dexamethasone and ABT-199 treatment of 3D MSC.

Figure 5. A comparative costs analysis between 3D ex vivo and in vivo preclinical models show an approximate 30% costs benefit and consequently increased financial sustainability when using ex vivo organoid models.

Study limitations

Here we use primary human MSC, which indeed may be difficult to source, and which may lead to donor variability. Using iPSC derived MSC to co-culture patient-derived leukaemia¹² may be a more suitable alternative, and would furthermore enable assay standardisation, thereby improving data reproducibility. Moreover, although we show that MSC showed no adverse effect following dexamethasone and ABT-199 treatment, many anti-cancer treatments notably cytotoxic chemotherapy may cause direct toxicity to the MSC. However, this further emphasises the need to include key microenvironment cell types when conducting anti-cancer drug testing, especially given cancer treatments influence interactions between leukaemia and their surrounding niche, with many cytotoxic treatments affecting both leukaemia and BM cells in patients. Furthermore, we acknowledge that an ideal representation of 3D biomimetic human cell-based model would include capturing both mesenchymal and vascular BM components within an immune-responsive context.